Mureka Text Prompt
The unglamorous node that feeds prompts to Mureka Song and Instrumental
- instrumental
- lyrics
Mureka Text Prompt is the boring, essential front end of the pack: two big text boxes and two output wires. You type the music description and the lyrics, and it hands them to the generation nodes in the exact typed format those nodes demand. Every workflow built on Mureka Song or Mureka Instrumental starts here.
Inputs and outputs
instrumental(STRING, multiline) - describe the music: genre, mood, tempo, instrumentation. It gets wrapped into the MUREKA_INSTRUMENTAL type.lyrics(STRING, multiline) - the words to sing, wrapped into MUREKA_LYRICS.instrumental(MUREKA_INSTRUMENTAL) - output; wire it into Mureka Song or Mureka Instrumental.lyrics(MUREKA_LYRICS) - output; wire it into Mureka Song's lyrics input.
Both fields are multiline, which matters - lyrics are usually a few verses and a chorus, and you don't want a cramped single-line box. For an instrumental-only workflow you can leave lyrics empty; the node passes an empty lyrics object along, and you simply don't feed it to a Song node.
Why this node exists
Mureka Song doesn't take raw text - it takes typed MUREKA_INSTRUMENTAL and MUREKA_LYRICS values, so it can't mistake a style prompt for lyrics or vice versa. Text Prompt is the polite adapter that keeps your text boxes on the graph and editable instead of hardcoding prompt values inside the generation node. It's the same job CLIP Text Encode does for image workflows: turn your words into the type the next node speaks.
Installing
Same as the pack: ComfyUI Manager → Install Custom Nodes → search "Mureka API", or manually:
cd ComfyUI/custom_nodes
git clone https://github.com/leewinder/comfyui-mureka-api.git
No models, no dependencies, and no API calls - this node never touches the network. It's pure text plumbing.
Troubleshooting
Honestly, there's little to go wrong. If the generator says the prompt inputs are missing or untyped, check that you wired this node's outputs (not a raw STRING from somewhere else) into the generation node. And be specific in the prompt - music models respond to "upbeat synthwave, 120 BPM, driving bass and analog pads" far better than "a cool song."
Inputs (2)
| Name | Type | Default | Description |
|---|---|---|---|
| instrumental | STRING | — | |
| lyrics | STRING | — |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| instrumental | MUREKA_INSTRUMENTAL | — |
| lyrics | MUREKA_LYRICS | — |